The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill explicitly instructs the agent to retrieve sensitive API keys from the execution environment variables QINGHU_TOKEN and QHKIT_TOKEN. Accessing credentials stored in the environment is a high-privilege action that increases the risk of accidental exposure.
[COMMAND_EXECUTION]: Instructions suggest that the agent perform API calls using system tools like curl. This guidance may lead the agent to construct shell commands that interpolate user-supplied input, such as search keywords or product links, which poses a risk of command injection if not properly sanitized.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from 1688 product pages and search results, which can contain hidden instructions intended to manipulate agent behavior.
Ingestion points: Data ingested from 1688goods_keyword_search_global and 1688goods_detail_global tools.
Boundary markers: The instructions do not provide explicit boundary markers or delimiters to isolate external data from the agent's core instructions.
Capability inventory: The agent has network access capabilities via curl and store modification capabilities via the Xiaofeng ERP integration.
Sanitization: No specific requirements for the validation or sanitization of data retrieved from external sources are mentioned.
[EXTERNAL_DOWNLOADS]: The skill performs network requests to vendor-associated domains www.iqinghu.com and xfdyorder.zzbtool.com to manage merchant data and product listings. While these are primary functions of the skill, they involve the transmission of business data to external services.

qinghu-douyin-quick-listing